This invention relates to a radio apparatus equipped with a distortion compensating function. More particularly, the invention relates to a radio apparatus with a distortion compensating function in which the amplification characteristic of a transmission power amplifier is linearized to suppress non-linear distortion and reduce power leakage between adjacent channels.
Frequency resources have become strained in recent years and greater reliance is being placed upon highly efficient digital transmission in radio communication. In a case where multivalued amplitude modulation is applied to radio communication, an important technique used on the transmitting side is to linearize the amplification characteristic of a power amplifier to thereby suppress non-linear distortion and reduce power leakage between adjacent channels. In a case where an amplifier having inferior linearity is used and an attempt is made to improve power efficiency, a technique which compensates for the distortion produced by this amplifier is essential.
FIG. 71 is a block diagram showing an example of a transmitter in a conventional radio apparatus. A group of digital data sent from an audio CODEC (coder/decoder) 1 is applied to a TDMA unit 2, where the data group is subjected to burst processing and to processing for separating the digital data group into an I signal (in-phase) and a Q signal (quadrature phase). The I and Q signals are outputted to a DA converter 3 in an allocated time slot. The DA converter 3 converts the I and Q signals to analog baseband signals and enters or inputs the baseband signals into an orthogonal modulator 4. The latter multiplies the input I and Q signals (the transmission baseband signals) by a reference carrier wave and a signal phase-shifted from the reference carrier by 90.degree., respectively, adds the two products, thereby performing an orthogonal conversion and outputs the result. A frequency converter 5 mixes the orthogonally modulated signal and a local oscillation signal to perform a frequency conversion, and a transmission power amplifier 6 amplifies the power of the carrier wave outputted by the frequency converter 5 and broadcasts the amplified signal into space from an antenna 7.
In such a communication apparatus, the input/output characteristic of the transmission power amplifier is a non-linear characteristic indicated by the dashed line in FIG. 72A. Non-linear distortion is produced as a result of the non-linear characteristic, and a frequency spectrum centered about a transmission frequency f.sub.0 possesses side lobes, as indicated by the dashed line in FIG. 72B. This results in leakage into adjacent channels and causes interference between the adjacent channels.
A wide variety of analog techniques which compensate for the occurrence of distortion have been proposed. These include linear amplification by combination of C-class amplification (LINC), a feed-forward method, an analog Cartesian method, a polar loop method and a pre-distortion method using non-linear elements.
However, these methods result in increased band noise and difficulties in phase adjustment when feedback gain is raised to provide further improvement upon distortion.
With this as a background, a method of compensating for distortion using a digital signal processing technique has been realized to cope with the greatly increased processing speed of digital signal processors (DSP) that has been made possible by recent advances in LSI techniques. Many technical papers dealing with digital methods of compensating for non-linear distortion have been published and the theory thereof is known. For example, see "Adaptive Linearization Using Pre-distortion" (Michael Faulkner and Mats Johanson), delivered at Victoria College (Australia), and "Adaptive Linearization Using Pre-distortion Experimental Results", IEEE Transaction on Vehicular Technology, Vol. No. 2, May, 1994. If these digital techniques are put into practical use, the aforementioned drawbacks of the analog methods are solved.
FIG. 73 is a block diagram of a transmission apparatus having a function which compensates for non-linear distortion digitally using a DSP. A digital data group (modulating signal) sent from the audio CODEC 1 is subjected to burst processing in the TDMA 2 and is entered into an arithmetic/control unit 8, which is constituted by a DSP, in an allocated time slot. As shown in FIG. 74, the arithmetic/control unit 8 includes a number of functional blocks, namely a distortion compensation coefficient memory 8a which stores distortion compensation coefficients h(Pi) (i=0-1023) conforming to levels 0-1023 of the modulating signal, a pre-distortion unit 8b which applies distortion compensation processing (pre-distortion) to the modulating signal using the distortion compensation coefficients h(Pi) conforming to the modulated signal levels, and a distortion compensation coefficient arithmetic unit 8c for comparing the modulating signal with a demodulated signal, obtained by demodulation using an orthogonal detector, described later, and for calculating and updating the distortion compensation coefficients h(Pi) in such a manner that the difference between the modulated and demodulated signals becomes zero.
The arithmetic/control unit 8 applies pre-distortion processing to the modulating signal using the distortion compensation coefficients h(Pi) conforming to the levels of the modulating signal, effects a conversion to I and Q signals and enters these signals into the DA converter 3. The DA converter 3 converts the I and Q signals to analog baseband signals and enters the baseband signals into the orthogonal modulator 4. The latter multiplies the input I and Q signals by a reference carrier wave and a signal phase-shifted from the reference carrier by 90.degree., respectively, adds the two products, thereby performing an orthogonal conversion, and outputs the result. The frequency converter 5 mixes the orthogonally modulated signal and a local oscillation signal to perform a frequency conversion, and the transmission power amplifier 6 amplifies the power of the carrier wave outputted by the frequency converter 5 and broadcasts the amplified signal into space from the antenna 7. Part of the transmission signal enters a frequency converter 10 via a directional coupler 9. The converter 10 performs a frequency conversion and enters the results into an orthogonal detector 11. The orthogonal detector 11 multiplies the input signal by the reference carrier wave and by the signal phase-shifted from the reference carrier by 90.degree., thereby performing orthogonal detection and reproducing the baseband I, Q signals from the transmitting side, and enters these signals into an AD converter 12. The latter converts the input I, Q signals to digital signals and enters the digital signals into the arithmetic/control unit 8. The latter compares the modulating signal and the demodulated signal, which is obtained by the orthogonal detector 11, by an adaptive algorithm using the LMS method (the method of least mean squares), and calculates/updates the distortion compensation coefficients h(Pi) so as to null the difference between the compared signals. The modulated signal to be transmitted next is then subjected to pre-distortion processing using the updated distortion compensation coefficients h(Pi) and the processed signal is outputted. By thenceforth repeating this operation, non-linear distortion of the transmission power amplifier is suppressed to reduce leakage of power to adjacent channels.
FIG. 75 is a diagram for describing distortion compensation processing based upon an adaptive algorithm. Shown in FIG. 75 is a multiplier 15a for multiplying a modulating signal (an input baseband signal) x(t) by a distortion compensation coefficient h(P), a transmission power amplifier 15b having a distortion function f(p), a feedback loop 15c for feeding back an output signal y(t) from the transmission power amplifier 15b, an arithmetic unit 15d for calculating power p [=x(t).sup.2 ] of the modulating signal x(t), a distortion compensation coefficient memory 15e for storing distortion compensation coefficients conforming to each power of the modulating signal x(t) and for outputting distortion compensation coefficients conforming to power of a currently prevailing modulating signal x(t), a conjugate complex signal output unit 15f, a subtractor 15g for outputting a difference e(t) between the modulating signal x(t) and the feedback demodulated signal y(t), multipliers 15g, 15h, and a multiplier 15i for multiplying by a step size parameter .heart..
The arrangement described performs the following operations: EQU h.sub.n (p)=h.sub.n-1 (p)+.heart.e(t)u*(t) EQU e(t)=x(t)-y(t) EQU y(t)=h.sub.n-1 (p)x(t)f(p) EQU u(t)=x(t)f(p)=h*.sub.n-1 (p)y(t) EQU P=.vertline.x(t).vertline..sup.2
where x, y, f, h, u and e represent complex numbers and * represents a conjugate complex number.
By performing the processing shown above, h(p) is updated so as to minimize the difference e(t) between the modulating signal x(t) and the feedback demodulated signal y(t), whereby the distortion compensation coefficients eventually converge to the optimum distortion compensation coefficient values to compensate for distortion of the transmission power amplifier.
Thus, the principle of digitally compensating for non-linear distortion entails feeding back and detecting a carrier wave obtained by orthogonal modulation using a modulating signal, digitally converting the amplitudes of the modulating signal (the transmission baseband signal) and the feedback signal (feedback baseband signal) and comparing the same, and updating the distortion compensation coefficients in real time based upon the results of comparison. As a consequence, if an amplitude error develops owing to offsets of the orthogonal modulator and orthogonal detector, an error occurs in the arithmetic operations and appropriate compensation of distortion can no longer be performed. At the present time, the analog orthogonal modulator and orthogonal detector generally in use are constituted by Gilbert-cell differential amplifiers, as is well known. As a result, an offset develops owing to a variance in element performance and a change in ambient temperature. Moreover, the offset value is influenced by the offset fluctuation of a DC amplifier used after the burst signal is subjected to the digital-to-analog (D/A) conversion.
FIG. 76 illustrates a QPSK-modulated wave expressed in a complex plane. If offset is present in a case where the original amplitude is a, as shown in FIG. 76, the offset is superposed on the original amplitude, as a result of which the arithmetic/control unit erroneously identifies the amplitude as being b. Since the method of digitally compensating for non-linear distortion uses this amplitude component in the comparison operation, the erroneous recognition of the amplitude leads to the occurrence of an error in distortion compensation so that proper compensation of distortion can no longer be carried out. In other words, executing the processing for updating the distortion compensation coefficients before compensating for the offset is not only meaningless but rather has an adverse influence upon operation. Thus, a technique to compensate for the offset is required in a radio apparatus having a distortion compensating function.
Further, the feedback baseband signals develop a phase shift with respect to the transmission baseband signals owing to the length of the transmission line or the devices in the transmission loop or feedback loop. When such a phase shift occurs, the original compensation for distortion can no longer be carried out accurately. A method of eliminating this phase shift is to compensate for the amount of the phase shift by changing the length of the transmission line along the way. However, this method is not practical because the phase shift varies depending upon the circuit mounting position or frequency. Accordingly, a technique which corrects for phase difference is required in a communication apparatus having a distortion compensating function.
Further, in the method of digitally compensating for non-linear distortion, the arithmetic/control unit constituted by the DSP executes distortion compensation processing digitally, as set forth above. The distortion compensation processing therefore is limited by the operating speed of the DSP. This means that if a data transmission is performed at high speed, the distortion compensation processing cannot keep pace. More specifically, in a case where the data transmission rate is on the order of 8 Kbps, there is enough time for the DSP to perform calculations in real time. When the transmission is performed at a high rate of 32 Kbps, however, the DSP cannot keep pace. This drawback becomes a fatal flaw in high-speed transmission, where the band becomes broader and the necessity for compensation of distortion becomes greater the higher the transmission rate. In other words, the conventional radio apparatus shown in FIG. 73 is not applicable to high-speed transmission.
In a radio apparatus of the TDMA type adapted to solve this problem, it has been contemplated to have the apparatus accumulate data (modulated signal data) transmitted in its own allocated time slot and perform the operations for distortion compensation even in time slots other than its own time slot. With this method, however, the distortion compensation coefficients are updated every burst. This means that the distortion compensation coefficient of each level is updated once in one TDMA frame period at best. Consequently, it takes time for the distortion compensation coefficients to converge. In the meantime, distortion compensation cannot be performed accurately, the band widens, interference between adjacent channels occurs and satisfactory data communication cannot take place. It should be noted that the radio apparatus of FIG. 73 cannot be applied to high-speed data transmission but does update the distortion compensation coefficients in real time. If data of identical levels enter n times in one time slot, the distortion compensation coefficients can be updated n times and the distortion compensation coefficients converge to the optimum value in a short time.
Thus, there is demand for a radio apparatus with a distortion compensating function, which apparatus is applicable to high-speed data transmission and causes the distortion compensation coefficients to converge to the optimum value in a short time.